The art of printing images with micro-fluid technology is relatively well known. A conventional permanent or semi-permanent ejection head has access to a local or remote supply of fluid. The fluid ejects from an ejection zone of the head to a print media in a pattern corresponding to pixels of images being printed. Over time, the heads and fluid drops have become smaller.
As part of recent trends, manufacturers increasingly have placed their inkjet ejection chips on ceramic substrates. Ceramics are relatively high modulus materials offering low coefficients of thermal expansion (CTEs). They are known to minimize chip bow in comparison to dies mounted directly on molded plastic substrates, especially when using epoxy die bond materials having elevated cure temperatures (typically in the range 110.degree.-150.degree. C.). In multi-chip, wide-swath ejection heads, ceramics provide benefit during thermal processing steps subsequent to preliminary die-to-substrate attachment, such as by maintaining a die's relative position during encapsulation cure, printed circuit board (PCB) attachment cure, wire bonding, etc.
Notwithstanding these advantages, ceramics have known drawbacks. One, manufacturers need to set dimensional tolerances fairly high since ceramics shrink during firing processes. As it stands, placement accuracies are compromised for fluid (ink) vias and bridges and their locations relative to each other and to components residing near terminal edges. Dry-pressed ceramics have typical tolerances of +/−200 .mu.m while those from more expensive ceramic injection molding (CIM) processes are smaller. Two, ceramics based in alumina have limited thermal conductivity and most do not incorporate any electrical functionality unless founded in tape cast varieties, such as HTCC (94-98% alumina) and LTCC (.about.40% glass in alumina). Tape cast varieties, however, are relatively expensive to fabricate and each comes with challenges in selecting compatibility relative to other materials. For instance, one grade of LTCC material examined by the inventors caused ink to flocculate, while selected HTCC materials required correspondingly low-conductivity trace (metal) materials, such as tungsten, when utilized in high-temperature firing environments. Three, there exists a practical limitation in the sizes of substrates that can be fabricated due to corresponding limitations in the sizes of modern presses. Naturally, this creates problems for manufacturers seeking to increase dimensions in printing swaths and chip arrays.
With reference to FIG. 1A, an ejection head 10 is formed with a PCB 12 and flexible cable 14. The PCB embodies a four wiring layer board and mounts on a ceramic substrate 16. The board provides electrical connections to ejection chips that reside in cutout “pockets” 18. The design adds electrical functionality over earlier, single layer TAB circuit, flexible circuit, solutions due to its presence of ground and power planes in the four wiring layers and an ability to combine and cross signal lines. When the PCB is configured with a material set of FR4 (the international grade designation for “Flame Retardant” (FR) fiberglass reinforced epoxy laminates), the design has further advantage in its compatibility with certain ink sets. However, there remains limits and unresolved process challenges as will be seen.
For example, a depth of the pocket 18 cannot be thicker than the chip it carries or material of the board will encroach into a paper gap distance of the ejection device. Also, the thickness of the PCB is best situated to remain .about.100 um lower than the chip surface to help minimize wire bond loop heights above the head's nozzle plate. However, a chip thickness of 450 .mu.m limits the thickness of the board to no more than 350 .mu.m (.about.14 mils) (e.g., 450 .mu.m-100 .mu.m). In a four (4) wiring layer board (e.g., wire bond pads on top, internal power and ground planes, routing interconnections and solder pads for flexible cable interconnect on bottom), this keeps the thickness of the board critically close to a minimum thickness of a board that can be made. It also limits the space to add a protective FR4 layer to the bottom of the board, such as over trace areas so that the only exposed metal on the bottom is for flexible cable attachment pads. By allowing a thicker die, on the other hand, the board thickness can increase but at an adverse cost to the substrate of poorer thermal dissipation.
In other setbacks, current corrosion protection for wiring traces is provided by an adhesive that attaches the board 12 to the ceramic substrate 16. However, since the boards are thin and flexible, they tend to warp after fabrication. This not only presents challenges for attaching the board to the ceramic, but compromises corrosion protection for the traces. Still other problems associated with board-to-ceramic attachment include: 1) squeezing epoxy excesses into the chip pocket causing, thereby interfering with later die mounting or contributing to volume variability in the pocket and making encapsulation heights unpredictable; or 2) creating interfacial voids that allow ink to access the delicate wiring traces on the bottom of the board.
Alignment or registration between the board and ceramic has also resulted in manufacturing concerns. One, the board warping makes it difficult to properly register components. Two, routing tolerances on the board provide little room to adjust components. Three, vision system pick-and-place fixtures require holding one piece steady while attaching the other with adhesive. As ejection heads move beyond single- or double-chip heads to larger arrays, the weight of now larger ceramics becomes problematic for smaller pick-and-place tools. Then, once placed, the board and ceramic must be fast-tacked during a preliminary cure step for transfer to a more permanent cure. However, UV curing lamps and other gluing fixtures are difficult to incorporate into standard tooling. Also, selections of compatible tack-and-hold and permanent bonding adhesives need be contemplated when making tooling selections. While none of the problems are individually insurmountable, a need exists in the art to simplify the process.
As part of solutions to the foregoing problems, third parties have introduced various epoxy based PCB's. They are thought to provide electrical functionality in low cost manufacturing items. They also sidestep many of the ceramic limitations noted above. Some have even been used as substrates for micro-fluidic applications. For example, U.S. Pat. No. 6,821,819 proposes using a PCB as a substrate and fluidic manifold for a microfluidic device chip like a biosensor, chemical sensor, or other electro microfluidic device. In U.S. Pat. No. 7,347,533, a piezoelectric inkjet printhead is constructed from a PCB material with driver chips affixed to the board. Neither of these, however, concern themselves with the challenges associated with precision methods needed to attach silicon inkjet chips to PCB materials. Rather, they relate to using conventional PCB material sets, which are insufficient for modern concerns.
When ejection head designs use a thermally cured adhesive, chip bow is affected by the modulus, CTE and thickness of the die, substrate, and adhesive together with the glass transition temperatures of these materials and the delta T between cure temperature and ambient. Using high Tg substrates with low CTEs are the better product for less stress on the die. Typical FR4 PCB materials, however, have CTEs of .about.16-20 ppm/.degree. C. and represent poor substrates for attachment to silicon chips having CTEs of 2-4 ppm/.degree. C. While ceramic materials with CTEs of 6-8 ppm/.degree. C. are a better match, they suffer the problems identified above. Also, standard FR4 based circuit boards may not provide enough rigidity/stiffness for micro-fluid applications. In turn, mounting a board assembly to a printhead body can induce significant stresses that may translate into deformation where the die interfaces with the FR4 substrate. Further, thermal conductivities of standard FR4 materials are very low when compared to ceramic (0.3-0.4 W/mK vs. .about.25 W/mK), and this could prove challenging for dissipating heat generated during printing.
In other modern designs, page-wide ejection heads are contemplated having arrays of very narrow ejection chips (<2 mm). Each chip includes multiple minute through holes feeding ink to each firing chamber instead of a single large via feeding ink. In comparison to larger conventional chips, the design enjoys more efficient use of silicon since large “real estate” need not be consumed by structurally supporting the large via or providing fluidic channels to each firing chamber. Instead, fluidic channels are confined to manifolds above and/or below the layers of patterned silicon on the chip. For one of many competing proposals on how best to construct the manifold, see, e.g., U.S. patent application Ser. No. 12/624,078, filed Nov. 23, 2009, incorporated herein by reference.
With reference to FIG. 4, an initial concept is seen for establishing fluidic and electrical interconnections in the '078 application. It includes an ejection chip and fluid manifold (silicon) 20. The ejection chip and manifold may be fabricated in a single piece of silicon or they may be separate pieces of silicon bonded in a laminar construct. The fluid manifold portion of this construct consists of fluid channels running the long axis of the chip, typically one per color of ink or per row of nozzles. The chip wire bonds 22 electrically to a printed circuit board 24. The board connects to a flexible circuit 26 which connects to printer electronics (not shown). Fluidically, the manifold communicates with a tile 28. The tile is silicon-based (absent thin film patterned layers) and has a thickness of about 400-600 .mu.m. Its manufacturing and interface to the manifold is more thoroughly described in U.S. patent application Ser. No. 12/568,739, filed Sep. 29, 2009, and such is also incorporated herein by reference. In general, however, the tile has ports on its top that fluidly mate with slots on the bottom of the manifold. It also has lateral channels at its bottom (orthogonal to the long axis of the ejection chip) that fluidly connect in stagger to through holes in a rigid substrate 30 (e.g. ceramic base). The substrate, in turn, provides both mechanical support and rigidity to the overall assembly. Ultimately, the goal of the fluidic arrangement is to fan-out the fluidic channels downward from the chip and condense them into a single port connection for each color. However, the design needs a “large enough” separation/seal distance so that a fluidic feed tube or plastic housing can be easily connected with a compliant rubber gasket or a needle-dispensed liquid adhesive. For adhesives typically used in micro-fluidic connections, the minimum achievable seal distances with needle dispense, without having either leaks, and crosstalk, between colors or excessive adhesive “squeeze-out” causing blockages in fluidic paths has been established as approximately 500 .mu.m of “land area” between adjacent fluidic features. Unfortunately, the printed circuit board 24 and ceramic base substrate 30 cause the design to encounter much of the same problems earlier described.
Accordingly, a further need exists in the art to accommodate pluralities of components in ejection heads having diverse functionality. Additional benefits and alternatives are also sought when devising solutions.